4 ms·
They normally release a 35b dense and an 27b moe (4B active per token) For context 35B on my m4 runs at 10 tokens a second, 27B moe runs 50-60 tokens a second.
by vorticalbox 1mo ago
They normally release a 35b dense and an 27b moe (4B active per token)
For context 35B on my m4 runs at 10 tokens a second, 27B moe runs 50-60 tokens a second.
- cpburns2009 1mo agoYou have your numbers switched. 27B is the dense model and runs slowly on unified memory. 35B (A3B active) runs great on unified memory.
- yencabulator 1mo ago27B dense or 35B-A3B MoE. You might be confusing it with Gemma 4 that has a 26B-A4B variant.